Heterogeneous Graph Propagation Network

Houye Ji, Xiao Wang, Chuan Shi, Bai Wang, Philip S. Yu

Abstract: Graph neural network (GNN), as a powerful graph representation technique based on deep learning, has shown superior performance and attracted considerable research interest. Recently, some works attempt to generalize GNN to heterogeneous graph which contains different types of nodes and links. Heterogeneous graph neural networks (HeteGNNs) usually follow two steps: aggregate neighbors via single meta-path and then aggregate rich semantics via multiple meta-paths. However, we discover an important semantic confusion phenomenon in HeteGNNs, i.e., with the growth of model depth, the learned node embeddings become indistinguishable, leading to the performance degradation of HeteGNNs. We explain semantic confusion by theoretically deriving that HeteGNNs and multiple meta-paths based random walk are essentially equivalent. Following the theoretical analysis, we propose a novel H eterogeneous graph P ropagation N etwork (HPN) to alleviate the semantic confusion. Specifically, the semantic propagation mechanism improves the node-level aggregating process via absorbing node's local semantic with a proper weight, which makes HPN capture the characteristics of each node and learn distinguishable node embedding with deeper HeteGNN architecture. Then, the semantic fusion mechanism is designed to learn the importance of meta-path and fuse them judiciously. Extensive experimental results show the superior performance of the proposed HPN over the state-of-the-arts.